Strange things on GBit / 1000->100 / net.inet.tcp.inflight.*

Andre Oppermann andre at freebsd.org
Fri Sep 17 14:30:35 PDT 2004


Andre Oppermann wrote:
> Raphael H. Becker wrote:
> 
>> On Fri, Sep 17, 2004 at 02:48:21PM +0200, Andre Oppermann wrote:
>>
>>> "Raphael H. Becker" wrote:
>>>
>>>> My guess: The 5.3-Boxes send bigger TCP-Windows than our switch has
>>>> buffer for each port resulting in massive packetloss or something like
>>>> that. The sender is "too fast" for the switch or the switch isn't able
>>>> to convert from 1000MBit to 100MBit under heavy load
>>>> (store&forward-buffer)
>>>
>>>
>>> Could you send me the output of (after you have run the 1000->100 test):
>>>
>>> # sysctl net.inet.tcp
>>> # sysctl net.inet.tcp.hostcache.list
>>> # netstat -s -p tcp
>>> # netstat -s -p ip
>>
>>
>> http://rhb.uugrn.org/FreeBSD/bugs/5.x/1000-to-100/check_before.txt
>> http://rhb.uugrn.org/FreeBSD/bugs/5.x/1000-to-100/check_after.txt
> 
> 
> Could you please provide the same information (except the second sysctl 
> one)
> from the target machine as well.  You don't have to show the difference
> between before and after but try to make sure that not much other traffic
> was going than the test.

And could you do this on the 5.3 machine:

  # sysctl net.inet.tcp.rfc3042=0

then restart and run the tests plus providing data for the 5.3
and 4.10 machine again.

-- 
Andre



More information about the freebsd-current mailing list